Proxies for Programmatic Ad Teams: Captures You Can Audit

Verify campaigns by geography with captures someone can audit: exit and page geo, creative, placement, landing URL and timestamp, one session per check.

If you verify that campaigns serve the right creative to the right geography, every check has to produce a capture someone can audit later: where the request came from, what ad appeared and where, where it led, and when. A proxy supplies the "where from". The rest of the record is what turns a page load into evidence a trafficker, publisher or client can act on.

This guide covers verification of your own campaigns and of placements you are entitled to check. Don't generate clicks on live ads you don't own; for your own campaigns, agree a test procedure with the ad network so verification traffic is excluded from billing.

What a verification capture must record

For example, a check of a Spanish mobile campaign might be stored as a single record like this one (the IDs, addresses and result are illustrative):

{
  "check_id": "camp2291-es-mobile-0007",
  "campaign_id": "2291",
  "expected_geo": "ES",
  "exit_ip": "203.0.113.24",
  "exit_country": "ES",
  "page_geo_signal": "es-ES consent banner, EUR prices",
  "placement_url": "https://publisher.example/noticias/articulo-123",
  "slot_id": "div-ad-top",
  "creative_id": "cr-88412",
  "creative_sha256": "9f2c...e1",
  "declared_click_url": "https://ads.network.example/click?c=88412",
  "landing_url_final": "https://advertiser.example/en/offer",
  "landing_language": "en",
  "user_agent": "Mozilla/5.0 (iPhone ...)",
  "viewport": "390x844",
  "captured_at": "2026-10-05T11:42:09Z",
  "screenshot": "captures/2291/0007.png",
  "result": "fail: Spanish creative landed on the English offer page"
}

Each field answers a question someone will ask:

  • Geo, three ways. expected_geo is the plan, exit_country is the network origin, and page_geo_signal is what the publisher page itself showed. If the last two disagree, the check is inconclusive, not a pass.
  • Creative. An ID from the ad markup plus a hash of the rendered asset, so "wrong creative" is provable even after the campaign rotates assets.
  • Placement. The page URL and slot, so the publisher can find it.
  • Landing URL. The final URL after the redirect chain, resolved from the declared click URL. In the example record, that is where the failure sits: a Spanish ad sending users to an English page.
  • Timestamp and device. Ads change through the day and by device class; a capture without both cannot be reproduced.
  • Screenshot. Visual proof for the cases a JSON field cannot settle.

Classify failures by type (wrong geo, wrong creative, missing ad, broken or wrong-language landing page, redirect to an unexpected domain) and report them per campaign and per placement. A single "pass rate" hides which party has to fix what.

Choosing exits: residential versus datacenter

Ad platforms commonly filter known datacenter traffic as invalid, so a page loaded through a datacenter address may show no ad, a house ad, or a different auction result from the one a real visitor sees. For checking what users in a country are served, residential proxies are usually the right tool: exits from a partner network in 160+ countries, with country and city segments in the username. Datacenter exits still fit checks that don't depend on the auction, such as confirming that landing pages, redirects and geo-specific offer pages load correctly from the United States, Italy and Spain, where Node4's datacenter IP blocks are.

Confirm the country you got, not just the one you asked for. Both gateways refuse a location they cannot serve rather than serving it from somewhere else: over an HTTP proxy connection the refusal is a 502 with an X-Node4-Country: unavailable header (a browser shows it only as a failed tunnel), and over SOCKS5 it is a "host unreachable" reply. On a successful HTTPS check, though, no country header reaches your browser, so record the exit address and check it independently, and store the page's own geo signal as the final word.

One check, one session

A browser opens many connections to render one page: the publisher, the ad server, the creative host, the tracker, the landing page. Rotating gateways choose a new exit for each HTTPS tunnel (and each plain-HTTP request), so a browser using a rotating gateway with no session ID can load the page from one exit and the ad from another. That is a broken check, even if every request succeeded.

Give every check its own session ID, and use it for the whole browser context:

curl -x "http://USERNAME-country-es-session-v0007:PASSWORD@gw-residential.node4.io:8082" https://api.ipify.org

How long the session holds depends on the product. On Rotating Datacenter and Rotating Unmetered plans, a session ID holds one exit for up to 30 minutes from first assignment, regardless of activity. Residential sessions are best effort, with a 30-minute ceiling, and can end sooner if the device behind the exit goes offline. A verification check takes well under a minute, so either is enough, provided each check starts a fresh browser context with fresh cookies and a fresh session ID. Reusing a context carries frequency caps and retargeting state into the next check. Puppeteer and Playwright proxies shows how to bind credentials per browser context.

Sizing the test matrix

The matrix is geographies times device classes times placements times however many repeats you need to see creative rotation. Write it down before buying anything; it sets both your concurrency and your transfer.

Browser captures pull full pages, images and video, so transfer per check is far higher than a plain HTTP fetch. Residential is billed per GB (residential prices); measure a day's matrix on a small sample before choosing a tier. For landing-page sweeps in the three datacenter countries, Rotating Datacenter is also billed per GB (rotating prices).

Where to start

Pick one campaign, three geographies and one device class. Run five checks per cell by hand in a normal browser on a local connection where you can, then through the proxy, and compare creative IDs and landing URLs. The ad verification proxy guide compares providers for the same job, and ad verification with proxies covers scheduling recurring checks. For the retail side of regional checks, see the market research guide; if your security team needs to test geographic access rules on your own sites, the cybersecurity and pentest guide covers recording which egress was used for what.